How to solve import error for pandas?

问题:

I installed Anaconda with python 2.7.7.
However, whenever I run "import pandas" I get the error:
"ImportError: C extension: y not built. If you want to import pandas from the source directory, you may need to run 'python setup.py build_ext --inplace' to build the C extensions first."
I tried running the suggested command but it stated that

skipping 'pandas\index.c' Cython extension (up-to-date)      
skipping 'pandas\src\period.c' Cython extension (up-to-date) 
skipping 'pandas\algos.c' Cython extension (up-to-date)      
skipping 'pandas\lib.c' Cython extension (up-to-date)        
skipping 'pandas\tslib.c' Cython extension (up-to-date)      
skipping 'pandas\parser.c' Cython extension (up-to-date)     
skipping 'pandas\hashtable.c' Cython extension (up-to-date)  
skipping 'pandas\src\sparse.c' Cython extension (up-to-date) 
skipping 'pandas\src\testing.c' Cython extension (up-to-date)
skipping 'pandas\msgpack.cpp' Cython extension (up-to-date)

Has anyone encountered this before and found a solution?

回答1:

Pandas has portions of its code written in C to make it run faster. If you tried to install pandas manually you would need to build it. Try reinstalling it with miniconda package manager here: http://conda.pydata.org/miniconda.html

and then you can just do

conda install pandas

There are very simple instructions on how to do it in the link below. Just do ctrl-f miniconda to find the section that talks about it

http://pandas.pydata.org/pandas-docs/dev/install.html

回答2:

I was having the same problem now with Python 3.4.3.

I was using pandas-0.18.0.

Upgrading (using pip) solved the issue for me:

[sudo] pip install --upgrade pandas

The final result of the upgrade:

Successfully installed numpy-1.13.3 pandas-0.21.0 python-dateutil-2.6.1 pytz-2017.3 six-1.11.0

After this, the issue was gone!

回答3:

I had the same problem and the issue came from an encoding problem. My os was previously set up in French and everything was fine. But then when I switched to English I had the error above.

You can type

locale

in the terminal to check the local environment variables.

When set up in French, I had this configuration: French config. Then, after I switched to English, I had: English config.

I then added the following lines in the .bash_profile under /Users/myName and everything went back to normal.

export LC_ALL=en_US.UTF-8
export LANG=en_US.UTF-8

回答4:

I was having this problem with python 2.7.13 here is my solution: 1. install Cython with

pip install Cython

2. install g++ and gcc

apt-get install gcc, g++

3. uninstall pandas

pip uninstall pandas

4. reinstall pandas

pip install pandas

then everything will be OK.

回答5:

I was unable to upgrade pandas with regular

pip install --upgrade pandas 

"tensorflow 1.6.0 has requirement numpy>=1.13.3, but you'll have numpy 1.13.1 which is incompatible."

However bumping it with:

pip install --upgrade pandas --force

solved issue completely

回答6:

I tried all the solutions above, but nothing works out...

Error Message

I got an error message with ipython

ImportError: C extension: iNaT not built. If you want to import pandas 
from the source directory, 
you may need to run 'python setup.py build_ext --inplace --force' 
to build the C extensions first.

and it suggests

$ python setup.py build_ext --inplace --force

Solution

My suggestion: Be careful about the version issue!

I clone pandas from the official github repo, then build it myself and install by pip

Following is the command I typed in terminal

$ cd pandas

$ python setup.py build_ext --inplace --force

$ sudo pip install .  # don't forget the dot 

or, if you want to install in your personal Linux account instead of under the system (due to multiple users issue)

you can add --user flag

$ pip --user install . # don't forget the dot, too

Now, everything works fine on my laptop

My configuration

Ubuntu 16.04
Python 2.7
Numpy 1.13.1 

Good luck!

回答7:

Actually, none of these answers worked for me in the following environment:

docker-compose # multiple containers, the managing one based on debian
Python 2.7
Django 1.8.19
numpy==1.11.3 # pinned to version, because of https://github.com/rbgirshick/py-faster-rcnn/issues/481

... more requirements

The following solution worked, after reading

https://github.com/pandas-dev/pandas/issues/18281

and

https://github.com/pandas-dev/pandas/issues/16715

which both addressed interim solutions and later recommended upgrading,

so I integrated into the Dockerfile

pip install -r requirements.txt \
&& pip install \
pandas==0.21.0 \
--force-reinstall \
--upgrade \
--no-deps \
--no-cache \
--find-links https://3f23b170c54c2533c070-1c8a9b3114517dc5fe17b7c3f8c63a43.ssl.cf2.rackcdn.com/ \
--no-index

which is mentioned in https://github.com/pandas-dev/pandas/issues/16715#issuecomment-310063504

I tried all solutions mentioned here, except the accepted answer, also because a) I don't want anaconda in a web production environment and b) it's not a good answer to foster frameworks or cli-solutions for architectures, where a package is not used standalone...

Furthermore, I dislike @colo's answer being downvoted, because it actually is a feasible solution in a certain environment.

For anyone finding this thread with similar requirements and expectations like me, I hope to have saved some minutes.

回答8:

try

/miniconda3/bin/conda install python

python: 3.6.0-0 --> 3.6.1-2

and

/miniconda3/bin/conda install pandas

Try the same with your Anaconda version.

回答9:

Instead of installing it with conda or pip, try to install it with your package manager:

sudo apt-get install python3-pandas

回答10:

I had this issue when I needed up upgrade from Python 32 bit to 64 bit to use tensorflow.

Running this command uninstalled pandas 0.21 and reinstalled 0.22 :

pip install --upgrade pandas

Sorted.

回答11:

I just had exactly the same issue when running tox.

Steps to solve:

1. Update setup.py to contain pandas==0.23.0 (instead of 0.21.0).
2. Remove .tox directory
3. Run tox again.